Resolving the genetic structure of striped marlin, Kajikia audax, in … · 2013. 11. 26. · Resolving the genetic structure of striped marlin, Kajikia audax, in the Pacific Ocean

Resolving the genetic structure of striped marlin,Kajikia audax, in the Pacific Ocean through spatialand temporal sampling of adult and immature fish

Catherine M. Purcell and Suzanne Edmands

Abstract: Spatial genetic structure in the highly migratory striped marlin (Kajikia audax) was examined using nuclear (mi-crosatellite) and mitochondrial (control region sequences) DNA markers. While previous studies on striped marlin were lim-ited by sampling design and sample size, this study employed a multiyear concurrent sampling scheme to collect tissuefrom seven locations representative of the species’ range in the Pacific: Japan, Hawaii, Southern California, Mexico, CentralAmerica, New Zealand, and Australia. Mature and immature specimens were analyzed separately to evaluate life-stage-specific population structure and movements. Microsatellite and sequence results revealed small but significant overall spa-tial subdivision (FST = 0.0145 and KST = 0.06995, respectively). Pairwise microsatellite analyses (n = 1199) revealed fourgroups: (1) Japan – Immature Hawaii – Southern California, (2) Mature Hawaii, (3) Mexico – Central America, and (4) Aus-tralia – New Zealand. Mitochondrial sequence analysis (n = 451) showed similar patterns; however, no significant differen-tiation was found between groups 1 and 2. This enhanced resolution of geographic genetic structure is important forunderstanding the complex migration patterns in this species. Moreover, the consistency among independent genetic studieson striped marlin provides strong support for management of at least three clearly delineated Pacific stocks.

Résumé : Des marqueurs nucléaires (microsatellites) et mitochondriaux (séquences de la région de contrôle) d’ADN nousont servi à déterminer la structure génétique spatiale du makaire strié (Kajikia audax), une espèce fortement migratrice.Alors que les études antérieures sur le makaire strié ont été limitées par leur plan d’échantillonnage et la taille des échantil-lons, notre étude utilise une méthode d’échantillonnage concurrent sur plusieurs années pour récolter des tissus dans sept ré-gions représentatives de l’aire de répartition de l’espèce dans le Pacifique, soit le Japon, Hawaii, le sud de la Californie, leMexique, l’Amérique Centrale, la Nouvelle-Zélande et l’Australie. Nous avons analysé les spécimens matures et immaturesséparément de manière à évaluer la structure génétique et les déplacements spécifiques au stade de vie. Les résultats desanalyses des microsatellites et des séquences indiquent globalement une subdivision faible mais significative (respective-ment, FST = 0,0145 et KST = 0,06995). Des analyses appariées des microsatellites (n = 1199) révèlent l’existence de quatregroupes : (1) Japon – immatures d’Hawaii – sud de la Californie, (2) matures d’Hawaii, (3) Mexique – Amérique Centraleet (4) Australie – Nouvelle-Zélande. L’analyse des séquences mitochondriales (n = 451) montre des patrons similaires; iln’existe cependant pas de différenciation significative entre les groupes 1 et 2. Cette différenciation améliorée de la structuregénétique géographique est importante pour comprendre les patrons complexes de migration de l’espèce. De plus, la concor-dance entre les études génétiques du makaire strié appuient fortement une gestion basée sur au moins trois stocks bien défi-nis dans le Pacifique.

[Traduit par la Rédaction]

Introduction

Large pelagic species such as tuna and billfish roam theworld’s oceans, free of any obvious physical barriers. It hasgenerally been thought that the highly migratory lifestyles ofthese marine fishes would lead to genetic homogeneityamong conspecific populations, yet molecular data show thatthis is not always true. While some of these free-roaming fish

show no genetic structure between or within oceans (e.g.,skipjack tuna (Katsuwonus pelamis) and wahoo (Acantho-cybium solandri) (Ely et al. 2005; Theisen et al. 2008), otherspecies exhibit varying levels of genetic subdivision, such asblue marlin (Makaira nigricans) (Finnerty and Block 1992;Buonaccorsi and Graves 2000; Buonaccorsi et al. 2001) andbigeye tuna (Thunnus obesus) (Durand et al. 2005), that showgenetic differentiation between Atlantic and Pacific Oceanpopulations. Genetic subdivision has also been seen withinocean basins; swordfish (Xiphias gladius) (Block and Reeb2000), Pacific sailfish (Istiophorus platypterus) (Graves andMcDowell 1995), and striped marlin (Kajikia audax) (Gravesand McDowell 1994; McDowell and Graves 2008) all re-vealed subdivision within the Pacific Ocean.While generally considered a bycatch species, striped mar-

lin are sold commercially in Japan, Taiwan, Australia, andHawaii, with smaller targeted fisheries in several locations.Striped marlin are also important to recreational fisheriesaround the Pacific, particularly in Hawaii, New Zealand, Aus-

Received 7 October 2010. Accepted 13 April 2011. Published atwww.nrcresearchpress.com/cjfas on xx October 2011.J2011-0097

Paper handled by Associate Editor Paloma Moran.

C.M. Purcell and S. Edmands. Department of BiologicalSciences, University of Southern California, Los Angeles, CA90089-0371, USA.

Corresponding author: Catherine M. Purcell (e-mail: [email protected]).

1861

Can. J. Fish. Aquat. Sci. 68: 1861–1875 (2011) doi:10.1139/F2011-104 Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

tralia, Southern California, and Mexico. These recreationalfisheries contribute greatly to coastal economies throughsportfishing and sportfishing-related tourism (Bromhead et al.2004). Given this species’ economical importance and thesigns of fishing-related population strain already occurring(Bromhead et al. 2004; Worm et al. 2005), it is important tounderstand its spatial distribution to develop effective man-agement strategies. This study focuses on examining the spa-tial genetic structure of striped marlin in the Pacific Ocean.

Previous workStriped marlin occurs in the Pacific and Indian oceans and

is considered the most abundant and widely distributed bill-fish (Nakamura 1985). In the Pacific, its distribution createsa horseshoe-shaped pattern across the ocean basin (Fig. 1),occurring in tropical, subtropical, and temperate regions. In-dividual striped marlin are capable of moving throughouttheir range as evidenced by tagging studies; however, thenorthernmost and southernmost extensions of their range areseasonal, as the waters become too cool during wintermonths (Squire 1972).Several stock scenarios have been proposed for striped

marlin: a single panmictic population within the Pacific (Sho-mura 1980), eastern–western Pacific stocks (Morrow 1957),and northern–southern Pacific stocks (Kamimura and Honma1958). Alternatively, a regional stock hypothesis is bolsteredby spatial and temporal partitioning of spawning events andmay also fit with the results of Graves and McDowell(1994), who used mitochondrial restriction fragment lengthpolymorphisms (RFLPs) in four Pacific locations to findshallow but significant levels of genetic subdivision, with 4/6significant pairwise comparisons. These findings were againsupported by a study using five microsatellites and mitochon-

drial control region sequences (McDowell and Graves 2008);the results suggested four regional stocks: Southwest Pacific(Australia), North Pacific (Japan–Taiwan–Hawaii–California),Mexico, and Southeast Pacific (Ecuador).Despite previous investigations into the striped marlin’s

population structure, questions remained regarding the ap-plicability of these studies to management issues because ofsampling design, temporal replicates, and number of speci-mens. McDowell and Graves (2008) attempted to addressthese issues from their first study (Graves and McDowell1994); however, with the complexities associated with sam-pling large pelagic fish, some of these difficulties persisted.Sample size, for instance, may have continued to be a limita-tion to the 2008 study; while 371 specimens were screenedwith microsatellites, only 85 specimens (from six of theseven locations) were sequenced at the mitochondrial controlregion. Pairwise mtDNA estimates did not detect genetic sub-division, possibly because of small sample size. These sam-pling limitations make it difficult to apply results of previousstudies to striped marlin management issues.In this project, sampling was conducted over seven loca-

tions representative of the striped marlin’s range in the Pa-cific, with concurrently sampled temporal replicates in six ofthe locations. Unlike previous billfish studies, to better assesswhere breeding populations occur, initial analyses kept ma-ture and immature individuals (in applicable locations) sepa-rate to determine if they were significantly different. Toincrease resolution power, two classes of molecular markerswere used: microsatellites (12 loci) and mitochondrial DNAcontrol region sequences. Concordance among markers addsconfidence in the findings; however, differences amongmarkers can also be informative and can help distinguish ge-netic signals due to selection or other forces.

Fig. 1. The Pacific Ocean range of striped marlin (Kajikia audax) (shaded area), sampling locations, and number of specimens collected inthose locations.

1862 Can. J. Fish. Aquat. Sci. Vol. 68, 2011

Published by NRC Research Press

Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

Given that genetic variability in striped marlin is high andthat sample size can greatly influence the patterns of geneticsubdivision, a total of 1199 specimens were screened withmicrosatellites and 451 specimens were sequenced at themtDNA control region. A greater number of loci, larger sam-ple sizes, and a concurrent sampling scheme may more con-clusively resolve population subdivision than previous studiesand increase the usefulness of these results to managingagencies. Additionally, temporal analyses of the same dataset (Purcell 2009; C. Purcell and S. Edmands, unpublisheddata) show that the spatial subdivision patterns reported hereare largely robust to temporal variance.While genetic subdivision is difficult to detect in migratory

species, as even a few migrants would erase patterns of sub-division, the variation observed in striped marlin suggestsdifferences in life-history traits and (or) population demo-graphics between this species and the other migratory speciesnot exhibiting genetic structure. Trying to understand thecause of these differences is crucial in developing effectivemanagement strategies to preserve this fishery.

Materials and methods

Sampling strategyFrom 2001 to 2007, samples were collected from seven lo-

cations chosen to be representative of the species’ range inthe Pacific: Japan (2003), Hawaii (2003, 2004, 2005), South-ern California (2001, 2002, 2004, 2005, 2006, 2007), Mexico(2003, 2004, 2005, 2006, 2007), Central America (2003,2004, 2005, 2006), New Zealand (2004, 2006), and Australia(2004, 2005) (Fig. 1). The samples were provided throughcommercial and recreational fishing efforts. Commercial sam-ples were collected through sponsored observer programs incollaboration with the National Marine Fisheries Service(NMFS), the Inter-American Tropical Tuna Commission(IATTC), the National Research Institute of the Far SeasFisheries (NRIFSF, Japan), and the Secretariat for the PacificCommunity (SPC, New Caledonia). Recreational sampleswere collected through independent environmental researchfirms, such as the Pfleger Institute of Environmental Re-search (Oceanside, California), Marine Conservation ScienceInstitute (Fallbrook, California), Pepperell Research and Con-sulting (New South Wales, Australia), and Nelson ResourcesConsulting (Miami, Florida), and scientists working with In-terdisciplinario de Ciencias Marinas (CICIMAR, Mexico).Recreational fishers also assisted in providing samples fromkill tournaments and from live fish on catch-and-release tripswith biopsy darts mounted on tagging poles. Sample num-bers varied by location because of the overall abundance ofstriped marlin in the area, the type of fishing used to obtainsamples, and the participation of contacts in those areas. Finor muscle tissue was preserved in ethanol or 20% dimethylsulfoxide buffer saturated with sodium chloride (Seutin et al.1991). Some samples were not used in genetic analyses be-cause of tissue degradation.

Sample preparationFor higher quality tissue, genomic DNA was extracted

from small amounts of tissue using Chelex (BioRad) or lysisreactions (Edmands et al. 2005). In the lysis extraction, a fewmuscle tissue fibers or a small amount of fin skin were added

to 50 µL lysis buffer at 65 °C for 1 h followed by 100 °C for15 min. For degraded samples, an overnight cetyltrimethyl-ammonium bromide – proteinase K incubation was followedby a standard phenol–chloroform extraction, with an ethanolprecipitation, and if necessary a lithium chloride wash.

Determination of maturitySamples were divided into reproductively immature and

mature individuals using several methods because of variablebiological information collected from the sample sources.Based on fish in the Coral Sea (Hanamoto 1977), 29 kg wasused as the mass at first maturity and 143 cm was used aseye fork length at first maturity. A factor of 1.2 (dressedmass × 1.2 = round mass) was used to convert dressedmasses into round masses based on published InternationalConvention for the Conservation of Atlantic Tunas (ICCAT)estimates for billfish (Mejuto et al. 2002). Lower jaw forklengths were converted into masses based on Kopf et al.(2005).

Microsatellite assaysTwelve microsatellite loci were used in this study: ten de-

veloped specifically for this project as described by Purcell etal. (2009) and two developed by Buonaccorsi and Graves(2000). Polymerase chain reaction (PCR) amplification con-ditions varied between microsatellite primer sets. Five micro-satellites were amplified using specific fluorescently labeledforward primers; the other seven microsatellites were ampli-fied with a modified nonlabeled forward primer containing a25 bp zip-code tag (Chen et al. 2000). Fluorescent comple-mentary primers for the zip-code tags were used for amplifi-cation of those modified microsatellites, with the resultingfragment sizes 25 bp longer. Both sets of fluorescent primersused Beckman WellRED D2, D3, or D4 dyes. PCR was con-ducted on a MJ Research PTC-200 DNA Engine and an Ap-plied Biosystems GeneAmp PCR System 9700 with thefollowing conditions: 15 ng template DNA, 0.25–1.0 µmol·L–1 primers, 1.5–3.5 mmol·L–1 MgCl2,0.25 mmol·L–1 dNTPs, 10 mmol·L–1 Tris-HCl, 50 mmol·L–1

KCl, and 0.3 U Taq polymerase in 12 µL total volume. Twocycling conditions were used: one for specifically labeled for-ward primers and another for labeled zip-code tags, describedin Purcell et al. (2009). PCR products were analyzed usingthe fragment analysis on a Beckman–Coulter CEQ 8000Capillary Sequencer and scored visually. Approximately 7%of samples were rerun for consistency in PCR amplificationand fragment analysis on the sequencer. For scoring consis-tency of microsatellite fragments, approximately 20% of sam-ples were rescored.

Mitochondrial control region assaysThe mitochondrial control region was amplified for 451

striped marlin specimens from the locations listed above,with five specimens from Ecuador used in two of the analy-ses. PCR was conducted using three universal primers (K: 5′-AGCTCAGCGCCAGAGCGCCGGTCTTGTAAA-3′) (Lee etal. 1995), (L19: 5′-CCACTAGCTCCCAAAGCTA-3′) (Ber-natchez et al. 1992), and (12 SAR-H: 5′-ATAGTGGGG-TATCTAATCCCAGTT-3′) (Palumbi et al. 1991) to amplifyapproximately 1000 bp of this region. PCR was conductedwith the following conditions: 25 ng template DNA,

Purcell and Edmands 1863


Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

1 µmol·L–1 forward and reverse primers, 0.25 mmol·L–1

dNTPs, 2.0–2.5 mmol·L–1 MgCl2, 10 mmol·L–1 Tris–HCl,50 mmol·L–1 KCl, and 1 U Taq polymerase in a 34 µL totalvolume. The cycling conditions consisted of the following:initial denaturation at 93 °C for 5 min, 35 cycles of denatu-ration at 93 °C for 1 min, primer annealing at 54 °C for30 s, followed by extension at 72 °C for 1 min 15 s. After35 cycles, a 5 min final extension at 72 °C was used, fol-lowed by a hold at 8 °C. After checking for amplificationon 1.2% agarose gels, successful amplifications were submit-ted to High-Throughput Sequencing Solutions operated bythe University of Washington, Department of Genome Sci-ences (Seattle, Washington) for ExoSAP PCR clean-up andsequencing using the primers from the original amplifica-tion. The entire control region was sequenced in both direc-tions using those primers. Sequences were examined andaligned using SEQUENCHER 4.6 (Gene Codes Corporation).

Microsatellite data analysesTo explore spatial patterns in striped marlin, the 1199

specimens were first arranged by collection location andthen further separated into mature and immature fish withineach location (where applicable). For each population, ob-served (Ho) and expected (He) levels of heterozygosity werecalculated using ARLEQUIN 3.1 (Excoffier et al. 2005). De-viations from Hardy–Weinberg equilibrium (HWE) were de-tected using FIS, and genotypic disequilibrium wascalculated for each locus-pair in each population usingGENEPOP 1.2 (Raymond and Rousset 1995) with 10 000 de-memorization steps, 1000 batches, and 10 000 iterations. Theprogram ML-NullFreq (Kalinowski et al. 2006) was used tocheck for null allele frequency in each locus and population.The original data set was corrected for null alleles based onestimated frequencies found using ML-NullFreq, and the ex-pected number of null homozygotes and heterozygotes werecalculated for all loci within each population. The expectednumbers of null homozygotes were added to individualswith missing data using the null allele “999”. Null heterozy-gotes were incorporated by randomly adding the null allele toexisting non-null allele homozygotes. The data incorporatingthe null alleles were permutated to mimic a round of sexualreproduction, thereby randomly mixing the null allelesthroughout the population using GENETIX 4.04 (Belkir etal. 2000).Using null-corrected data, the genetic structure of mature

and immature groups within locations was determined. If nosignificant differences were found between mature and imma-ture samples in a location, they were combined for the re-maining analyses. However, if groups showed significantstructure, then they remained as separate groups for the restof the analyses.Because the null allele correction method permutated data

within locations to shuffle the null alleles, the original (non-corrected) data from the sample groups determined abovewere used to estimate observed and expected heterozygosity,FIS, and genotypic disequilibrium using the programs and pa-rameters mentioned above. Genetic variability among–withinstriped marlin populations, measured by allelic richness (AR)

and the number of alleles (NA), was also calculated with thenon-null-corrected data using FSTAT 2.9.3 (Goudet 1995).Weir and Cockerham’s overall FST and pairwise FST values

were calculated for all sample groups using 10 000 permuta-tions in GENETIX 4.04 (Belkir et al. 2000), followed by astraight Bonferroni correction of the pairwise estimates. Forcomparison, this was conducted for null-corrected and non-corrected data. Hedrick’s G′ST (2005) was calculated from es-timates of Nei and Chesser’s (1983) GST and HS given inGENETIX (Belkir et al. 2000) for the null-corrected data.The program ISOLDE within GENEPOP was used to testthe correlation between geographic distance, measured as theaverage distance among sampling locations in kilometres (km)and population structure, as estimated with FST. Correlationsignificance between genetic structure and geographical dis-tance was assessed using the Mantel test in GENEPOP with1000 permutations.The model-based Bayesian clustering program STRUC-

TURE 2.2 (Pritchard et al. 2000) was also used to examinepopulation structure. Five replicates were run with a burn-inof 100 000 steps and 500 000 Markov chain Monte Carlo(MCMC) steps for K from 1 to 11, using the admixturemodel and the assumption that allele frequencies are corre-lated among populations.

Mitochondrial data analysesThe number of haplotypes (h), haplotype diversity (hd),

and average number of differences between sequences (K)were calculated for sequences in each location usingDNASP 4.50.3 (Rozas et al. 2003). The haplotype-based sta-tistic (HST) and nucleotide sequence-based statistic (KST)were calculated in DNASP. Population pairwise comparisonswere conducted for the average number of pairwise nucleo-tide differences (Kxy) and for the nearest-neighbor statistic,Snn, which is a measure of how often the most similar se-quences are from the same location (Hudson 2000) inDNASP. Also in DNASP, pairwise genetic subdivision esti-mates were calculated using the sequence-based statistic(KST) according to Hudson et al. (1992). A neighbor-joiningtree was created using ClustalX 2.0.10 (Larkin et al. 2007)and MEGA 4 (Tamura et al. 2007) using a bootstrap testwith 5000 replicates. A hierarchical structuring of the mito-chondrial sequences using an analysis of molecular variance(AMOVA) within ARLEQUIN was used to assess the rela-tive contribution of variance among groups, within groups,and within populations using the distance matrix of pairwisedifferences among sequences.

Results

Microsatellite analysesDuring the microsatellite analysis, initial comparisons of

observed and expected heterozygosities, FIS values, and nullallele frequency estimates indicated that a null allele correc-tion should be incorporated into the data set (SupplementalTable S11, available online). Null frequencies ranged from0% to 33.6% of homozygotes for a few locus–location combi-nations; however, it is important to note that even higher es-timates may have reflected only one or two homozygous

1Supplementary data are available with the article through the journal Web site at http://nrcresearchpress.com/doi/suppl/10.1139/f2011-104.



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

individuals. Following this correction, the data were used tocompare the mature and immature specimens within the loca-tions. Based on FST estimates, immature and mature Mexicanspecimens and immature and mature Central American speci-mens were grouped together for the remaining analyses.However, significant structure was detected between imma-ture and mature Hawaiian specimens, and they were thereforemaintained as separate groups.

Genetic variability (microsatellites)Because the null allele correction permuted data within lo-

cations to randomly distribute null alleles among samples, theoriginal data (not corrected) was used in all summary statis-tics. Genetic variability (Table 1) within striped marlin popu-lations varied widely among loci; allelic richness ranged from2.4 (locus 164) to 7 (locus Mn08), both in Japan, but no sig-nificant differences were found among locations in allelicrichness (p = 0.99, using a two-tailed test implemented inFSTAT with 1000 permutations). The number of alleles var-ied widely depending on the locus, from 2 alleles in Australiaand New Zealand (locus 218) to 49 alleles (locus Mn08) inMature Hawaii and Mexico, but again was not significantlydifferent among locations (p = 0.54, using a two-tailed testimplemented in FSTAT with 1000 permutations).Observed levels of heterozygosity (Ho) also varied greatly

among microsatellite loci (Table 1); however, average hetero-zygosity across all loci for each location fell within a narrowrange. Lowest average Ho (0.652) was in Southern California,and the highest averages (0.727 and 0.696) were in Australiaand New Zealand, respectively; neither Ho nor He among lo-cations were significantly different (p = 0.98 and p = 0.99,respectively). Significant heterozygote deficits were seen inMature Hawaii (seven loci), Immature Hawaii (five loci),Australia (four loci), Mexico (three loci), New Zealand (twoloci), and Southern California and Central America (one lo-cus each). Japan showed no significant differences.FIS was also used to investigate deviations from HWE (Ta-

ble 1), and tests were conducted for the eight populationgroups, including the two groups from Hawaii. Numbersranged from positive values, indicating a deficit of heterozy-gotes, to negative values, indicating an excess of heterozy-gotes. Significant deviations were found in Japan (threeloci), Mature Hawaii (nine loci), Immature Hawaii (fourloci), Southern California (four loci), Mexico (four loci),Central America (three loci), New Zealand (one locus), andAustralia (three loci) (Table 1). FIS values were predomi-nately positive, and 32 of the 33 significant values were pos-itive, indicating a heterozygote deficit.Genotypic disequilibrium for all locus-pairs was calculated

for the eight sample groups (Supplemental Table S21) usingthe log likelihood ratio statistic (G test) and the Markov chainmethod with the parameters 10 000 dememorization steps,1000 batches, and 10 000 iterations in GENEPOP. FollowingBonferroni correction, significant disequilibrium was detectedin 37 locus-pairs in Australia and one locus-pair in MatureHawaii (Supplemental Table S21). In total, 56% of locus-pairsin Australia showed significant genotypic disequilibrium.

Genetic structure (microsatellites)Spatial structure was calculated using Wright’s FST accord-

ing to Weir and Cockerham (1984) and Hedrick’s standar-

dized genetic differentiation estimate, G′ST, at all loci foreach population using GENETIX. Significance levels wereassessed using 10 000 permutations. These estimates wereconducted for both null-corrected and original data to exam-ine effects of null alleles. Although FST values were small,they were significant for the null-corrected and original datasets, 0.0145 and 0.0123, respectively. As expected, Hedrick’sG′ST estimates for both groups were much larger than the FSTestimates, 0.0802 (null-corrected data) and 0.0559 (originaldata).Pairwise FST comparisons for the null-corrected data

(Table 2) detected significant subdivision in 23 of 28 com-parisons. The five sample pairs without significant differen-tiation were Japan – Immature Hawaii, Japan – SouthernCalifornia, Immature Hawaii – Southern California, Mexico –Central America, and New Zealand – Australia. The valuesand range of values in pairwise comparisons were larger us-ing G′ST rather than FST estimates, with the largest value ofG′ST, 0.1345, found between Southern California and Aus-tralia, and the smallest, 0.0050, found between Japan andSouthern California (Table 2). This is compared with thelargest FST estimate, 0.0374, between Mexico and Australia,and the smallest, 0.0013, between Southern California andJapan (Table 2). Despite the spatial heterogeneity detected us-ing the estimates above, analyses using STRUCTURE showedthat a single population (K = 1) provided the best fit to thedata with the largest log-likelihood estimate of –41 364.2.Pairwise FST comparisons of the original data revealed

similar patterns to the null-corrected analysis, but with fewersignificant relationships (Table 2). Key differences were thatthe original data showed no significant pairwise differencesbetween Mature Hawaii and the Immature Hawaii, SouthernCalifornia, and New Zealand sample groups. The smallestFST value, 0.0002, was found between Japan and ImmatureHawaii, and the largest, 0.0340, between Central Americaand New Zealand. Hedrick’s G′ST showed a wider range inestimates, from 0.0004 (Japan – Immature Hawaii) to 0.0771(Mexico–Australia). A significant positive correlation wasfound between geographic distance (average distance amongsampling locations) and population structure (FST) for stripedmarlin specimens from different locations (one-sided Manteltest with 1000 permutations, p = 0.0087, R = 0.3559). Thescatterplot of the average distance among sampling locations(km) and FST is shown (Fig. 2).

Genetic variability (mitochondrial sequences)A total of 451 individuals were sequenced (GenBank ac-

cession Nos. JF755428–JF755878) from eight locations: Ja-pan, Hawaii, Southern California, Mexico, Ecuador (n = 5,not shown in Table 3), Central America, New Zealand, andAustralia. Out of 451 control region sequences, 351 wereunique haplotypes. Relative to the number of sequences, thenumber of haplotypes in each location was very high, rang-ing from 18 (Central America) to 91 (Mexico) (Table 3).Not surprisingly, haplotype diversity was very high amongthe locations (Table 3). The average number of differencesbetween sequences (K) ranged from a low in Australia (26)to a high in Central America (42).

Genetic structure (mitochondrial sequences)The overall genetic differentiation based on the mitochon-



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

Table 1. For each location, data is shown for the number of specimens analyzed with each microsatellite locus (n), allelic richness (AR), number of alleles per locus (NA), observed (Ho)and expected (He) levels of heterozygosity (Ho/He), and FIS values used to detect deviation from the Hardy–Weinberg equilibrium.

Microsatellite locus

24 157 162 164 218 235 105 149 155 193 Mn01 Mn08

JP (n = 119)n 110 110 108 111 108 108 61 76 68 85 43 50AR 5.5 3 5.1 2.4 2.5 5.4 6.2 2.8 3.6 6.3 5.2 7.2NA 19 5 17 8 4 20 19 5 8 26 13 34Ho/He 0.882/0.881 0.491/0.558 0.759/0.817 0.396/0.383 0.481/0.510 0.870/0.869 0.902/0.923 0.400/0.443 0.448/0.599 0.906/0.925 0.744/0.821 0.898/0.968FIS –0.001 0.121 0.071 –0.036 0.057** –0.002* 0.023 0.098 0.254 0.021** 0.094 0.073

MT HW (n = 312)n 282 279 275 279 222 274 100 125 166 180 98 193AR 5.6 3.1 5.2 3.2 3 5.8 6.3 3.1 3 6.4 5.3 7NA 25 6 21 10 7 27 25 6 6 30 15 49Ho/He 0.876/0.889 0.513/0.560*** 0.763/0.832*** 0.446/0.510 0.489/0.547*** 0.813/0.888*** 0.788/0.920* 0.333/0.403 0.333/0.423*** 0.817/0.927 0.786/0.844 0.932/0.958***FIS 0.015 0.085** 0.084*** 0.126 0.107*** 0.084*** 0.145*** 0.173 0.212*** 0.119*** 0.069*** 0.027*

IM HW (n = 227)n 196 197 201 207 172 204 128 144 120 157 76 125AR 5.5 2.6 5.3 3.3 2.6 5.4 6.4 2.2 3.5 6.6 5.3 7.1NA 26 6 19 9 5 21 30 4 8 41 15 40Ho/He 0.857/0.880 0.538/0.552 0.735/0.826*** 0.398/0.489 0.444/0.504* 0.853/0.874 0.754/0.927*** 0.486/0.472 0.431/0.566*** 0.821/0.935** 0.803/0.844 0.920/0.960FIS 0.026 0.026 0.110*** 0.186 0.118** 0.024 0.187*** –0.03 0.240*** 0.123*** 0.049** 0.042

SC (n = 66)n 58 61 59 59 61 63 29 44 37 38 22 37AR 6.1 2.6 4.6 2.3 2.4 5.5 6.1 2.9 3.3 6.6 5.4 7NA 19 4 16 7 4 16 17 5 6 22 9 30Ho/He 0.862/0.911 0.508/0.535 0.746/0.751 0.356/0.353 0.574/0.517 0.857/0.875 0.862/0.910 0.349/0.436 0.286/0.462 0.868/0.934 0.773/0.855 0.779/0.966*FIS 0.054* 0.051 0.007 –0.01 –0.111 0.021* 0.053 0.202 0.385 0.071* 0.098 0.197***

MX (n = 239)n 218 218 214 224 183 211 84 106 117 188 79 174AR 5.5 4.6 2.1 3 6.3 3.2 6.6 2.8 5.4 2.3 4.5 7NA 19 7 16 6 6 22 24 5 7 35 13 49Ho/He 0.858/0.879 0.564/0.581 0.687/0.760 0.339/0.330 0.404/0.509*** 0.806/0.856*** 0.845/0.921 0.500/0.465 0.624/0.617 0.855/0.934 0.744/0.774 0.920/0.957*FIS 0.024 0.029 0.096 –0.027 0.205** 0.059*** 0.083** –0.076 –0.011 0.085** 0.039 0.04

CA (n = 105)n 100 99 104 103 93 102 31 60 48 68 37 64AR 5.6 4.9 2.3 3.4 5.9 3.9 6.5 2.3 5.2 2.9 4.7 7NA 17 6 16 7 3 16 16 7 9 28 9 37Ho/He 0.890/0.884 0.485/0.567 0.680/0.783 0.427/0.383 0.462/0.503 0.824/0.847 0.667/0.907 0.417/0.458 0.604/0.674 0.897/0.933 0.794/0.782 0.794/0.963*FIS –0.007 0.145 0.132 –0.115 0.081 0.027 0.269** 0.091 0.105 0.039* –0.015 0.177***

NZ (n = 86)n 76 71 78 79 80 79 56 29 67 58 56 67AR 5.7 2.8 5.3 3.7 2 5.6 6.4 3.2 2.8 6.3 5 6.9NA 17 6 13 7 2 15 23 5 6 23 12 33Ho/He 0.895/0.893 0.535/0.542 0.808/0.860 0.582/0.653 0.538/0.499 0.899/0.892 0.746/0.933* 0.517/0.542 0.349/0.403 0.772/0.920* 0.804/0.827 0.911/0.951FIS –0.002 0.012 0.062 0.109 –0.077 –0.008 0.203*** 0.047 0.137 0.162 0.029 0.043

1866Can.

J.Fish.

Aquat.

Sci.

Vol.68,

2011

Publishedby

NRCResearch

PressCan

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

drial sequences was calculated by two methods, HST and KST,that use the c2 statistical approach proposed by Hudson et al.(1992, eq. 2 and eq. 9, respectively). Both methods showedsignificant differentiation, with HST, the haplotype-based sta-tistic estimated at 0.00493 (p < 0.001) and KST, the nucleotidesequence-based statistic estimated at 0.06995 (p < 0.001).Population comparisons conducted for the average number ofpairwise nucleotide differences (Kxy) (Table 3) showed thelowest estimate of Kxy was 27.9 (Australia – New Zealand)and the highest was 44.3 (Central America – Southern Cali-fornia). For the nearest-neighbor statistic (Snn; Table 3), valueswere lowest between Immature Hawaii and Japan (0.55) andhighest between Australia and Mexico (0.95).Pairwise genetic subdivision patterns were explored using

the sequence-based statistic KST (Table 3). Becausesequence-based methods utilize information not only basedon the frequency of haplotypes, but also on the numbers ofdifferences between haplotypes, they are powerful in detect-ing structure in longer sequences or within smaller samplesizes (Hudson et al. 1992; Hudson 2000). With the KST esti-mates, no significant differentiation was detected betweenany pairwise combination of the Japan, Mature Hawaii, Im-mature Hawaii, or Southern California locations. Also nopopulation genetic structure was detected between Mexicoand Central America or between New Zealand and Australia(Table 3). The KST estimates ranged from –0.0026 (Japan –Immature Hawaii) to 0.1172 (Australia – Central America).A neighbor-joining haplotype tree (Fig. 3) was generated for

the mtDNA control region sequences. For values over 50%,replicate trees where individuals clustered together in the boot-strap test (5000 replicates) are shown next to the branches.However, branches where separations were reproduced lessthan 50% of the time were collapsed. This tree showed someof the same spatial patterns revealed by the pairwise KST val-ues, particularly clusters of individuals from Mexico, CentralAmerica, and Ecuador. However, other areas of the treeshowed mixed regions of sequences from different locations,reflecting the overall low values of genetic differentiation.An AMOVA of spatial variation was computed by the dis-

tance matrix using pairwise distances in ARLEQUIN (10 000permutations) for different groupings (Table 4), includingseveral previous striped marlin stock models: northern vs.southern Pacific, western vs. eastern Pacific, and regional(Central America and Ecuador were grouped together be-cause of the small sample size from Ecuador). Although, themajority of variation was contained within populations for allcomparisons, those particular groupings did not minimizevariation among populations within groups. The model thatmost minimized within-group variance, “Best fit” (Table 4),showed that although most variation (approximately 90%) oc-curred among samples within populations, about 10%(10.55%) of the variation was contained among groups, witheffectively no variation among populations within groups.This model contained three groups: Mature Hawaii – Imma-ture Hawaii – Japan – Southern California, New Zealand –Australia, and Mexico – Ecuador – Central America. Two al-ternate models were also provided for comparison. One wassimilar to the Best fit model except that Mature Hawaii wasmoved into a separate group, and while the AMOVA resultswere similar, slightly less variance was distributed amonggroups (9.77%). The second alternative moved Southern Cal-T

able

1(concluded).

Microsatellite

locus

24157

162

164

218

235

105

149

155

193

Mn01

Mn08

AU

(n=

45)

n42

4441

4543

4542

4444

4442

42AR

5.1

2.5

5.4

42

4.9

6.3

2.1

3.4

6.4

4.5

6.6

NA

144

139

29

213

620

822

Ho/He

0.857/0.838

0.636/0.556

0.675/0.852*

0.711/0.731

0.512/0.499

0.822/0.845

0.786/0.920*

0.545/0.484

0.614/0.656

0.750/0.926***

0.881/0.802

0.929/0.944**

FIS

–0.023

–0.146

0.210*

0.027

–0.026

0.027

0.147*

–0.129

0.066

0.192**

–0.1

0.017

Note:

Significantpvalues

aredenotedas

follo

ws:*,

p<

0.05;**,p

<0.01;***,

p<

0.001.

JP,Japan;MTHW,M

atureHaw

aii;IM

HW,ImmatureHaw

aii;SC

,SouthernCalifo

rnia;MX,M

exico;

CA,C

entral

America;

NZ,N

ewZealand;AU,A

ustralia.



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

ifornia into the group with the rest of the eastern Pacific;however, results revealed this model was not nearly as goodof a fit as the previous two groupings.

Discussion

Microsatellite analysesAllelic richness and number of alleles were variable, de-

pending on the locus and location. Heterozygosity levelswere high but similar to what has been reported for stripedmarlin in a previous study (McDowell and Graves 2008) andin other pelagic fish such as Atlantic bluefin tuna (Carlssonet al. 2004) and Atlantic bigeye tuna (Gonzalez et al. 2008).Null alleles, a common cause of discrepancies between ob-served and expected heterozygosities, are known to be prob-lematic and prevalent in microsatellite markers (Dakin and

Avise 2004; Hedgecock et al. 2004; Kalinowski et al. 2006).In this study, null alleles were detected in a subset of lociand populations. While often mentioned as a caveat of popu-lation genetic data, usually no further treatment of the nullalleles is attempted (Dakin and Avise 2004). However, giventhe detection of null alleles and the number of significant de-viations in the heterozygosity and FIS estimates, it was im-portant to correct for null alleles in this data set, as it didchange the number of significant pairwise relationships.Overall FST and G′ST values were low using microsatel-

lites, but significant and in the range reported for other pela-gic marine species (the reported median in Ward et al. 1994;O’Reilly et al. 2004; Rooker et al. 2007). Analysis of theeight groups of null-corrected samples revealed that stripedmarlin form four significantly different populations. The firstgroup located in the southwest Pacific contained Australiaand New Zealand. Australia had unusually high levels ofgenotypic disequilibrium, with 56% of locus-pairs showingsignificant disequilibrium, compared with only one otherlocus-pair (Mature Hawaii) in all other populations. One pos-sible explanation was misidentification of other billfish spe-cies within that region’s collection; however, all Australiansamples had a striped marlin maternal background after se-quencing the mitochondrial control region. Mexico and Cen-tral America in the eastern Pacific formed the second group;however, this group was significantly different from SouthernCalifornia, despite also being in the eastern Pacific. The ge-netic heterogeneity found between Southern California andMexico (primarily the Baja California region of Mexico) issurprising because of the close physical proximity betweenthose two locations; yet it is also supported in the findingsof McDowell and Graves (2008). Interesting spatial patternswere found in the northern Pacific. No significant differenceswere found between Japan and Southern California, but bothwere significantly different from Mature Hawaii. Interest-ingly, juvenile striped marlin caught around Hawaii were sig-

Table 2. Weir and Cockerham’s pairwise FST values (above diagonal) and Hedrick’s G′ST (below diagonal) for (i) null-correctedmicrosatellite data and (ii) original microsatellite data.

Population JP MT HW IM HW SC MX CA NZ AUNull-corrected dataJP 0.0074*** 0.0035 0.0013 0.0099*** 0.0076*** 0.0224*** 0.0284***MT HW 0.0289 0.0066*** 0.0077** 0.0193*** 0.0176*** 0.0090*** 0.0219***IM HW 0.0135 0.0263 0.0054 0.0154*** 0.0131*** 0.0121*** 0.0173***SC 0.005 0.0302 0.0196 0.0170*** 0.0144*** 0.0236*** 0.0371***MX 0.0337 0.0714 0.0539 0.0567 0.0026 0.0299*** 0.0374***CA 0.0273 0.0686 0.0483 0.0508 0.0087 0.0298*** 0.0337***NZ 0.0845 0.0358 0.0457 0.0862 0.1041 0.1103 0.0071AU 0.1064 0.0879 0.0655 0.1345 0.1298 0.1233 0.027

Original dataJP 0.0081*** 0.0002 0.0016 0.0070*** 0.0077*** 0.0216*** 0.0263***MT HW 0.0161 0.0029 0.003 0.0165*** 0.0219*** 0.0047 0.0183***IM HW 0.0004 0.006 0.0003 0.0120*** 0.0135*** 0.0092*** 0.0155***SC 0.0031 0.0057 0.0008 0.0111*** 0.0172*** 0.0151*** 0.0301***MX 0.0135 0.032 0.0236 0.0206 0.0004 0.0308*** 0.0377***CA 0.0149 0.0432 0.0272 0.0332 0.0008 0.0340*** 0.0335***NZ 0.044 0.0094 0.0191 0.03 0.0615 0.0699 0.0083AU 0.0549 0.0379 0.0327 0.0614 0.0771 0.0702 0.0177

Note: Significant p values are denoted as follows: **, p <0.01; ***, p < 0.001.

Fig. 2. Correlation between geographic distance (km) and popula-tion structure (FST) for striped marlin specimens (R = 0.3559), withsignificance determined by one-sided Mantel tests (p = 0.0087).



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

nificantly different from mature fish in that area, but not sig-nificantly different from Japan and Southern California fish.Thus, according to the null-corrected data, striped marlinfrom Japan, California, and Immature Hawaiian fish form thethird population, and Mature Hawaiian striped marlin consti-tute the fourth group.Despite the migratory behaviors that likely underlie the de-

tected genetic patterns, there is still significant correlation be-tween geographic distance and population structure,suggesting isolation by distance. Although some of these be-haviors indicate trans-Pacific movement (Japan – SouthernCalifornia), while others suggest limited mixing despite closeproximity (Mexico – Southern California), in general, it ap-pears that distance does limit genetic mixing, albeit atPacific-wide scales. In contrast with pairwise analyses, theprogram STRUCTURE did not support more than one popu-lation of striped marlin in the Pacific. However, assignmentmethods have limited power when genetic differentiation islow among populations (Manel et al. 2005; Saenz-Agudeloet al. 2009).When null allele frequencies were not incorporated into

the data, overall FST and G′ST estimates were lower, butmany pairwise relationships remained the same. The primarydifference was that Mature Hawaii showed fewer significantpairwise relationships; compared with null-corrected data nosignificant differences were detected with Southern Califor-nia, Immature Hawaii, or New Zealand. For the originaldata, the pairwise FST values that changed in significancehad some of the smallest significant values in the null-corrected data set. Thus, in this study, the null allele correc-tion provided additional power to detect significant relation-ships rather than change the trends of those relationships.Interestingly, changes in significant pairwise groupings all in-volved Mature Hawaii, a group that contained the highestnumber of deviations in observed heterozygosity and FIS esti-mates. This indicates that null allele correction was importantto spatial heterogeneity analyses.

Mitochondrial DNA sequence analysesMitochondrial control region sequences contained a large

number of haplotypes, and the haplotype diversity (hd) washigh but similar to other marine fish (Hauser et al. 2001;

Table 3. Based on mitochondrial control region sequence analyses, (i) diversity estimates; (ii) Kxy; (iii) Snn; and (iv) pairwiseKST estimates.

Location JP MT HW IM HW SC MX CA NZ AUDiversity estimatesns 43 92 34 25 131 20 53 48h 38 86 33 23 91 18 48 34hd 0.993 0.999 0.998 0.99 0.99 0.989 0.996 0.981K 33 34 29 30 39 42 30 26

Pairwise Kxy

MT HW 33.9IM HW 31.1 31.9SC 31.3 32.4 29.5MX 39.2 40.2 39.1 39.4CA 43.2 45.3 43.9 44.3 40.8NZ 33.2 33.5 30.6 32.2 39.9 44.1AU 31.7 32 28.8 30.9 38.9 43.5 27.9

Pairwise SnnMT HW 0.69**IM HW 0.55 0.62SC 0.57 0.69 0.61MX 0.92*** 0.86*** 0.89*** 0.81***CA 0.85*** 0.87*** 0.83*** 0.74** 0.79NZ 0.83*** 0.69*** 0.74** 0.83*** 0.92*** 0.81**AU 0.89*** 0.77*** 0.79*** 0.89*** 0.95*** 0.94*** 0.62*

Pairwise KST

MT HW 0.0025IM HW –0.0026 0.0004SC –0.0025 0.0048 –0.0006MX 0.0320*** 0.0478*** 0.0436*** 0.0374***CA 0.0642*** 0.0587*** 0.1029*** 0.1133*** 0.0025NZ 0.0272** 0.0183*** 0.0139* 0.0392*** 0.0587*** 0.0859***AU 0.0462*** 0.0352*** 0.0278** 0.0646*** 0.0742*** 0.1172*** –0.001

Note: Diversity estimates include the number of sequences (ns), number of haplotypes (h), haplotype diversity (hd), and average number ofdifferences between sequences (K); Kxy is the average number of pairwise nucleotide differences among sequences; Snn is the nearest-neighborstatistic; and pairwise KST estimates are based on the sequence-based statistics of Hudson et al. (1992). Significant p values are denoted asfollows: *, p < 0.05; **, p < 0.01; ***, p < 0.001.



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

Carlsson et al. 2004; Alvarado-Bremer et al. 2005). The east-ern Pacific locations, Mexico and Central America, share thecharacteristics of having both the highest average numbers ofdifferences between sequences (K), whereas Australia waslowest, which is perplexing given the unusual genotypic dis-equilibrium patterns in Australia mentioned above.The overall genetic structure was significant using HST and

KST estimates, and pairwise analyses showed patterns similarto the microsatellite analyses, with the same groups in theeastern Pacific (Mexico – Central America) and the south-west Pacific (Australia – New Zealand); however, there wasa change in the North Pacific group. Sequence analyses didnot reveal significant structure among Mature Hawaii and Ja-pan, Immature Hawaii, or Southern California, and thus thesespecimens formed one group in the North Pacific. Thenearest-neighbor statistic, Snn, supported the KST results.AMOVA comparisons also supported the mitochondrial

pairwise results, with the best fit resulting in three groupings:(1) Australia and New Zealand, (2) Japan, Immature Hawaii,Mature Hawaii, and Southern California, and (3) Mexico, Cen-tral America, and Ecuador. Interestingly, the northern–southernand eastern–western stock model groupings proved to be apoor fit with the sequence data. The similarly poor fit ofSouthern California with the rest of the eastern Pacific sequen-ces supports the pattern detected in the microsatellite analysis.

Overall spatial structureFollowing the analysis of nuclear microsatellites and mito-

chondrial control region sequences, the spatial distributionand underlying migration patterns of striped marlin in the Pa-cific are starting to be understood. Large numbers of devia-tions in heterozygosity and FIS values in mature Hawaiianspecimens were at least partially explained by the presence ofnull alleles, but may also be indicative of a Wahlund effect,

Fig. 3. Neighbor-joining haplotype tree of mitochondrial control region sequences. Branches corresponding to separations reproduced lessthan 50% of the time are collapsed. The percentages of replicate trees where individuals clustered together in the bootstrap test (5000 repli-cates) are shown next to the branches for values over 50%. Symbols represent individual sequences.



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

where sampling occurred across cohorts of juveniles or sub-groups (Johnson and Black 1984; Lenfant and Planes 2002),especially as this location may serve as a stepping stone orfeeding area for non-local striped marlin. There was disagree-ment between microsatellite and mitochondrial analyses; onepossible reason may be lower resolution power with the feweranalyzed mtDNA sequences compared with the number ofspecimens analyzed with microsatellites. Alternatively, micro-satellites may be more sensitive than mtDNA sequences, re-flecting very low levels of gene flow between locations(Feulner et al. 2004; Keeney et al. 2005). It is also possiblethat Japanese juveniles migrating to (or through) Hawaii maybe reflected in samples collected in that location, and mixingof these groups could impact analyses for both mature andimmature Hawaiian samples. Japanese juveniles around Ha-waii may grow and be included in analyses of mature Hawai-ian fish while still not reproductively contributing to thislocation. Similarly, although spawning in striped marlin isnow confirmed in Hawaii (Hyde et al. 2006), collections ofHawaiian juveniles likely have Japanese fish mixed in. Move-ment of juvenile striped marlin into Hawaiian waters is sup-

ported by previous reports (Squire and Suzuki 1990), wherethe fish likely use the area as a feeding ground before movingon to spawning locations (Matsumoto and Kazama 1974).Overall, the North Pacific shows very little spatial subdivi-

sion among the Japan, Immature Hawaii, Mature Hawaii, andSouthern California locations. The seasonal Southern Califor-nia location does not have a known spawning location, andstriped marlin in this area likely originate in Japan (Nishi-kawa et al. 1978). Fish from Japan may move eastward withthe Kuroshiro Current to feeding grounds near the HawaiianIslands; when currents or environmental conditions are right,some fish may continue eastward until they reach SouthernCalifornia. The seasonal population in Southern California isnot large, but their movement into the region is a regular oc-currence, although the timing and number of fish each yearvaries. While it is uncertain how long the fish remain in thisregion, they may eventually utilize the North Equatorial Cur-rent to move back across the Pacific to their spawninggrounds in Japan, or they may represent a population sink,not reproductively contributing to any population.Based on the genetic analyses, Mexican and Central Amer-

Table 4. Analysis of molecular variance (AMOVA) of spatial variation in striped marlin control region sequences.

Structure tested Variance % Total F statistics pWest–east (MX, EC, CA, SC) vs. (JP, NZ, AU) vs. (MT HW and IM HW)Among groups 0.792 Va 4.23 FCT = 0.0423 0.102Among populations within groups 0.951 Vb 5.08 FSC = 0.0530 <0.001Within populations 16.988 Vc 90.69 FST = 0.0931 <0.001

North–south (EC, AU, NZ) vs. (JP, SC, MA HW, IM HW, MX, CA)Among groups 0.084 Va 0.45 FCT = 0.0045 0.39Among populations within groups 1.562 Vb 8.38 FSC = 0.0842 <0.001Within populations 16.988 Vc 91.16 FST = 0.0884 <0.001

Separate populations (IM HW and MT HW) vs. (CA, EC) vs. (JP) vs. (AU) vs. (SC) vs. (NZ) vs. (MX)Among groups 2.338 Va 12.57 FCT = 0.1257 0.037Among populations within groups –0.730 Vb –3.92 FSC = –0.0449 <0.001Within populations 16.988 Vc 91.35 FST = 0.0865 <0.001

All populations together (AU, MT HW, IM HW, JP, MX, NZ, SC, CA, EC)Among groups 1.605 Va 8.63 FCT = N/A NAAmong populations within groups N/A N/A FSC = N/A NAWithin populations 16.988 Vb 91.37 FST = 0.0863 <0.001

Best fit (MT HW, IM HW, JP, SC) vs. (AU, NZ) vs. (MX, EC, CA)Among groups 2.001 Va 10.55 FCT = 0.1055 <0.001Among populations within groups –0.013 Vb –0.07 FSC = –0.0008 <0.001Within populations 16.988 Vc 89.52 FST = 0.1048 <0.001

Alternate group 1 (IM HW, JP, SC) vs. (MT HW) vs. (AU, NZ) vs. (MX, EC, CA)Among groups 1.831 Va 9.77 FCT = 0.0977 <0.001Among populations within groups –0.085 Vb –0.45 FSC = –0.0050 <0.001Within populations 16.989 Vc 90.68 FST = 0.0932 <0.001

Alternate group 2 (MT HW, IM HW, JP) vs. (MX, SC, EC, CA) vs. (AU, NZ)Among groups 1.266 Va 6.72 FCT = 0.0672 0.003Among populations within groups 0.576 Vb 3.06 FSC = 0.0328 <0.001Within populations 16.988 Vc 90.22 FST = 0.0979 <0.001

Note: Computed by the distance matrix using pairwise differences in Arlequin (10 000 permutations). AU, Australia;CA, Central America; EC, Ecuador; JP, Japan; IM HW, Immature Hawaii; MT HW, Mature Hawaii; MX, Mexico;NZ, New Zealand; SC, Southern California.



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

ican striped marlin appear to form one stock that is differen-tiated from other Pacific populations. Despite the genetic ex-clusion of Southern Californian striped marlin from the restof the eastern Pacific, tagging data indicate that striped mar-lin caught in Southern California move south into Baja Cali-fornia, Mexico, corresponding to cooling water temperaturesoff California (Domeier 2006). However, the Southern Cali-fornia fish do not appear in representative samples fromMexico; this may happen for a couple of reasons. First, theirbehavior in Mexico may reduce their catchability to the rec-reational fishery, which was the primary sample source inthat region. This is plausible because the spawning seasonfor the Mexican population coincides with the southwardmovement of Southern California fish, and evidence suggeststhat local striped marlin shift locations in preparation forspawning (Armas et al. 1999; Domeier 2006). Tournamentcatch rates indicate that the Mexican striped marlin popula-tion is far larger than the Southern California population. Itwould seem logical for recreational and commercial fishersto shift their effort to follow the larger, year-round populationof Mexican fish to their spawning grounds. As the stripedmarlin from Southern California do not appear to be spawn-ing in Mexico, they may largely escape the fishing pressureby not moving to the spawning locations. Alternatively, theSouthern Californian striped marlin may not stay long in thewaters off of Mexico, but rather move through Mexico to fol-low currents back across the Pacific. Through these (and pos-sibly other) behavioral modifications, and with the relativesize of the Southern California population compared withthe Mexico population, it is not surprising that they are notsignificantly sampled within this region.In the southwest Pacific, Australia and New Zealand form

another independent stock. Australia displayed unusual ge-netic variability patterns, with high numbers of locus-pairsin genotypic disequilibrium and heterozygosity deviations.One possibility is gene flow occurring with Indian Oceanstriped marlin on a semiregular basis, given the right environ-mental conditions (Bromhead et al. 2004), but with little ge-netic work done on Indian Ocean striped marlin, it is difficultto determine if mixing is driving these differences. If Pacific–Indian Ocean mixing was occurring, sequence variationwould be expected to be higher here than in other locations,and the results do not support this hypothesis. However, ifgene flow from the Indian Ocean was primarily male-mediated, then mitochondrial sequence variation would notbe greatly impacted. Hybridization with other regional bill-fish species is another possibility. All maternal backgroundswere striped marlin, but a unidirectional hybrid cross is pos-sible, though this has not been reported in billfish previously.Like Southern California, striped marlin in New Zealand areseasonal. The New Zealand fish likely originate from Aus-tralia and perhaps from other unsampled areas in theCentral–South Pacific (Bromhead et al. 2004); they are notedfor relatively high dispersal distances, likely correlated withchanging water temperatures, in contrast with Australianstriped marlin that make much shorter distance coastal shifts(Kopf et al. 2005; Domeier 2006).The presumed large-scale movements underlying the ge-

netic subdivision patterns of striped marlin during differentportions of their life history, particularly in the North Pacific,demonstrate the need for a concurrent, temporally, and spa-

tially representative sampling approach. While results of thisstudy were similar to McDowell and Graves (2008), they didnot analyze immature and mature fish separately; therefore nostructure was detected between those groups in Hawaii.McDowell and Graves (2008) also found structure betweenMexico and Ecuador, while in this study Mexico and CentralAmerica comprised a single stock. This difference is difficultto explain, as the larger numbers of markers and samplesused in this study should increase resolution. One possibleexplanation is that the more southern location, Ecuador,showed stronger differentiation than Central America in thisstudy; however, the collecting range for specimens from thesetwo locations were relatively close geographically. Alterna-tively, this may represent sampling artifact in McDowell andGraves (2008), caused by relatively small, single collectionsin those locations (Mexico: n = 32, 1994; Ecuador: n = 39,1995). Despite these differences, overall values of FST weresimilar (using microsatellites) between this study and theMcDowell and Graves 2008 study (0.0145 and 0.0130, re-spectively). This study also showed concordance between thetwo molecular markers, including the same significant pair-wise relationships (with the exception of mature Hawaiianspecimens). This suggests that small sample size was likelythe reason that McDowell and Graves (2008) did not detectsignificant subdivision in pairwise mtDNA estimates.

Conservation implicationsWhile it is not easy to subdivide migratory species into

distinctly manageable units, there are patterns from this studythat can be applied at a management level. Based on allspecimens, the north Pacific (Japan, Mature Hawaii, Imma-ture Hawaii, and Southern California) showed little geneticdifferentiation among locations. However, subtle populationstructure between mature Hawaiian specimens and the otherlocations in this group indicates that effective migration (mi-gration with reproductive contributions) between Hawaii andother North Pacific locations is probably limited. Therefore,it may be more accurate to consider them as separate stockseven though fishing efforts in Hawaii can impact both Hawai-ian and Japanese striped marlin populations. The Hawaiianlocation is clearly important to several populations withinthe North Pacific, and because of this connection, temporalanalyses may be very informative in managing this area.Next, Mexico and Central America appear differentiatedfrom the other Pacific locations. Despite the movement ofSouthern Californian fish through that region, the eastern Pa-cific should be managed as a separate stock. Finally, theSouthwest Pacific (Australia and New Zealand) form anotherdistinctive stock. Australia, in particular, may be impacted bymigration or hybridization in ways that are different fromother regions in the Pacific.Based on the analysis of over 1000 specimens collected

from seven locations, using two classes of molecularmarkers, striped marlin populations showed shallow but sig-nificant overall genetic subdivision with evidence for at leastthree (Australia – New Zealand, Japan – Southern Califor-nia – Hawaii, Mexico – Central America) and possibly four(Mature Hawaii) putative stocks in the Pacific. What may bemore complicated is the level of independence of thesegroups, particularly with the connections in the North Pacificand possibly the Australian – New Zealand group with poten-



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

tial ties to the Indian Ocean. However, for a pelagic speciescapable of long migrations, striped marlin show considerablespatial variation throughout their range in the Pacific.

AcknowledgementsWe thank D. Hedgecock and M. Hinton for valuable

manuscript suggestions; V. Pritchard and A. Vogel for techni-cal advice; V. Allain, D. Curran, L. Daccarett, M. Domeier,E. Everett, M. Hinton, D. Holts, R. Humphreys, J. Hyde,T. Kazama, R.K. Kopf, R. Nelson, P. Offield, S. Ortega-Garcia, J. Pepperell, H. Saito, K. Yokawa, and the ZaneGrey/Drambuie Tournament Staff for assistance in samplecollection. This work was supported by the Offield FamilyFoundation and the EPA STAR Fellowship FP91655901.

ReferencesAlvarado Bremer, J.R., Vinas, J., Mejuto, J., Ely, B., and Pla, C.

2005. Comparative phylogeography of Atlantic bluefin tuna andswordfish: the combined effects of vicariance, secondary contact,introgression, and population expansion on the regional phylo-genies of two highly migratory pelagic fishes. Mol. Phylogenet.Evol. 36(1): 169–187. doi:10.1016/j.ympev.2004.12.011. PMID:15904864.

Armas, R.G., Sosa-Nishizaki, O., Rodríquez, R.F., and Perez, V.A.L.1999. Confirmation of the spawning area of the striped marlin,Tetrapturus audax, in the so-called core area of the eastern tropicalPacific off Mexico. Fish. Oceanogr. 8(3): 238–242. doi:10.1046/j.1365-2419.1999.00102.x.

Belkir, K., Borsa, P., Chickhi, L., Raufaste, N., and Bonhomme, F.2000. GENETIX 4.04 Logiciel sous Windows TM, pour laGénétique des Populations. Laboratoire Génome, Populations,Interactions, Universite de Montpellier II, Montpellier, France.

Bernatchez, L., Guyomard, R., and Bonhomme, F. 1992. DNAsequence variation of the mitochondrial control region amonggeographically and morphologically remote European brown trout,Salmo trutta, populations. Mol. Ecol. 1(3): 161–173. doi:10.1111/j.1365-294X.1992.tb00172.x. PMID:1344992.

Block, B., and Reeb, C. 2000. Genetic analysis of populationstructure in Pacific swordfish (Xiphias gladius) using microsa-tellite DNA techniques. Annual Report for Fiscal Year 2000. JointInstitute for Marine and Atmospheric Research, Pelagic FisheriesResearch Division, Honolulu, Hawaii.

Bromhead, D., Pepperell, J., Wise, B., and Findlay, J. 2004. Stripedmarlin: biology and fisheries. Bureau of Rural Sciences, Canberra,Australia. ISBN: 0642475938.

Buonaccorsi, V., and Graves, J. 2000. Isolation and characterizationof novel polymorphic tetra-nucleotide microsatellite markers fromthe blue marlin, Makaira nigricans. Mol. Ecol. 9(6): 820–821.doi:10.1046/j.1365-294x.2000.00915-2.x. PMID:10849299.

Buonaccorsi, V.P., McDowell, J.R., and Graves, J.E. 2001.Reconciling patterns of inter-ocean molecular variance fromfour classes of molecular markers in blue marlin (Makairanigricans). Mol. Ecol. 10(5): 1179–1196. doi:10.1046/j.1365-294X.2001.01270.x. PMID:11380876.

Carlsson, J., McDowell, J.R., Diaz-Jaimes, P., Carlsson, J.E., Boles,S.B., Gold, J.R., and Graves, J.E. 2004. Microsatellite andmitochondrial DNA analyses of Atlantic bluefin tuna (Thunnusthynnus thynnus) population structure in the Mediterranean Sea.Mol. Ecol. 13(11): 3345–3356. doi:10.1111/j.1365-294X.2004.02336.x. PMID:15487994.

Chen, J., Iannone, M.A., Li, M.S., Taylor, J.D., Rivers, P., Nelsen, A.J., Slentz-Kesler, K.A., Roses, A., and Weiner, M.P. 2000. Amicrosphere-based assay for multiplexed single nucleotide poly-

morphism analysis using single base chain extension. GenomeRes. 10(4): 549–557. doi:10.1101/gr.10.4.549. PMID:10779497.

Dakin, E.E., and Avise, J.C. 2004. Microsatellite null alleles inparentage analysis. Heredity, 93(5): 504–509. doi:10.1038/sj.hdy.6800545. PMID:15292911.

Domeier, M.L. 2006. An analysis of Pacific striped marlin(Tetrapturus audax) horizontal movement patterns using pop-upsatellite archival tags. Bull. Mar. Sci. 79(3): 811–825.

Durand, J., Collet, A., Chow, S., Guinand, B., and Borsa, P. 2005.Nuclear and mitochondrial DNA markers indicate unidirectionalgene flow of Indo-Pacific to Atlantic bigeye tuna (Thunnusobesus) populations, and their admixture off southern Africa. Mar.Biol. 147(2): 313–322. doi:10.1007/s00227-005-1564-2.

Edmands, S., Feaman, H.V., Harrison, J.S., and Timmerman, C.C.2005. Genetic consequences of many generations of hybridizationbetween divergent copepod populations. J. Hered. 96(2): 114–123.doi:10.1093/jhered/esi014. PMID:15618307.

Ely, B., Vinas, J., Alvarado Bremer, J.R., Black, D., Lucas, L.,Covello, K., Labrie, A.V., and Thelen, E. 2005. Consequences ofthe historical demography on the global population structure oftwo highly migratory cosmopolitan marine fishes: the yellowfintuna (Thunnus albacares) and the skipjack tuna (Katsuwonuspelamis). BMC Evol. Biol. 5(1): 19. doi:10.1186/1471-2148-5-19.PMID:15725349.

Excoffier, L., Laval, G., and Schneider, S. 2005. Arlequin ver. 3.0: anintegrated software package for population genetics data analysis.Evol. Bioinform. Online, 1: 47–50.

Feulner, P.G.D., Bielfeldt, W., Zachos, F.E., Bradvarovic, J., Eckert,I., and Hartl, G.B. 2004. Mitochondrial DNA and microsatelliteanalyses of the genetic status of the presumed subspecies Cervuselaphus montanus (Carpathian red deer). Heredity, 93(3): 299–306. doi:10.1038/sj.hdy.6800504. PMID:15241451.

Finnerty, J., and Block, B. 1992. Direct sequencing of mitochondrialDNA detects highly divergent haplotypes in blue marlin (Makairanigricans). Mol. Mar. Biol. Biotechnol. 1(3): 206–214. PMID:1308204.

Gonzalez, E.G., Beerli, P., and Zardoya, R. 2008. Genetic structuringand migration patterns of Atlantic bigeye tuna, Thunnus obesus(Lowe, 1839). BMC Evol. Biol. 8(1): 252. doi:10.1186/1471-2148-8-252. PMID:18798987.

Goudet, J. 1995. FSTAT version 1.2: a computer program to calculateF statistics. J. Hered. 86(6): 485–486.

Graves, J.E., and McDowell, J.R. 1994. Genetic analysis of stripedmarlin (Tetrapturus audax) population structure in the Pacific.Can. J. Fish. Aquat. Sci. 51(8): 1762–1768. doi:10.1139/f94-177.

Graves, J.E., and McDowell, J.R. 1995. Inter-ocean geneticdivergence of istiophorid billfishes. Mar. Biol. 122(2): 193–203.

Hanamoto, E. 1977. Fishery oceanography of striped marlin. Part 2:Spawning activity of the fish in the Southern Coral Sea. SO-NipponSuisan Gakkaishi, Bull. Jpn. Soc. Sci. Fish. 43: 1279–1286.

Hauser, L., Turan, C., and Carvalho, G.R. 2001. Haplotype frequencydistribution and discriminatory power of two mtDNA fragments ina marine pelagic teleost (Atlantic herring, Clupea harengus).Heredity, 87(6): 621–630. doi:10.1046/j.1365-2540.2001.00956.x.PMID:11903557.

Hedgecock, D., Li, G., Hubert, S., Bucklin, K., and Ribes, V. 2004.Widespread null alleles and poor cross-species amplification ofmicrosatellite DNA loci cloned from the Pacific Oyster,Crassostrea gigas. J. Shellfish Res. 23(2): 379–385.

Hedrick, P.W. 2005. A standardized genetic differentiation measure.Evolution, 59(8): 1633–1638. PMID:16329237.

Hudson, R.R. 2000. A new statistic for detecting genetic differentia-tion. Genetics, 155(4): 2011–2014. PMID:10924493.

Hudson, R.R., Boos, D.D., and Kaplan, N.L. 1992. A statistical test



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

for detecting geographic subdivision. Mol. Biol. Evol. 9(1): 138–151. PMID:1552836.

Hyde, J.R., Humphreys, R., Musyl, M., Lynn, E., and Vetter, R. 2006.A central north Pacific spawning ground for striped marlin,Tetrapturus audax. Bull. Mar. Sci. 79(3): 683–690.

Johnson, M.S., and Black, R. 1984. The Wahlund effect and thegeographical scale of variation in the intertidal limpet Sipho-naria sp. Mar. Biol. 79(3): 295–302. doi:10.1007/BF00393261.

Kalinowski, S., Wagner, A., and Taper, M. 2006. ML-Relate: acomputer program for maximum likelihood estimation of related-ness and relationship. Mol. Ecol. Notes, 6(2): 576–579. doi:10.1111/j.1471-8286.2006.01256.x.

Kamimura, T., and Honma, M. 1958. A population study of the so-called makajiki (striped marlin) of both northern and southernhemispheres of the Pacific. Part I: Comparison of externalcharacters. Rep. Nankai Fish. Res. Lab. 8: 1–11. [Translated byI. Nakamura.]

Keeney, D.B., Heupel, M.R., Hueter, R.E., and Heist, E.J. 2005.Microsatellite and mitochondrial DNA analyses of the geneticstructure of blacktip shark (Carcharhinus limbatus) nurseries inthe northwestern Atlantic, Gulf of Mexico, and Caribbean Sea.Mol. Ecol. 14(7): 1911–1923. doi:10.1111/j.1365-294X.2005.02549.x. PMID:15910315.

Kopf, R.K., Davie, P.S., and Holdsworth, J.C. 2005. Size trends andpopulation characteristics of striped marlin, Tetrapturus audax,caught in the New Zealand recreational fishery. NZ J. Mar.Freshw. Res. 39(5): 1145–1156. doi:10.1080/00288330.2005.9517381.

Larkin, M.A., Blackshields, G., Brown, N.P., Chenna, R., McGettigan,P.A., McWilliam, H., Valentin, F., Wallace, I.M., Wilm, A., Lopez,R., Thompson, J.D., Gibson, T.J., and Higgins, D.G. 2007.Clustal W and Clustal X version 2.0. Bioinformatics, 23(21):2947–2948. doi:10.1093/bioinformatics/btm404. PMID:17846036.

Lee, W., Conroy, J., Howell, W., and Kocher, T. 1995. Structure andevolution of teleost mitochondrial control regions. J. Mol. Evol.41(1): 54–66. doi:10.1007/BF00174041. PMID:7608989.

Lenfant, P., and Planes, S. 2002. Temporal genetic changes betweencohorts in a natural population of a marine fish, Diplodus sargus.Biol. J. Linn. Soc. 76(1): 9–20. doi:10.1111/j.1095-8312.2002.tb01710.x.

Manel, S., Gaggiotti, O.E., and Waples, R.S. 2005. Assignmentmethods: matching biological questions with appropriate techni-ques. Trends Ecol. Evol. 20(3): 136–142. doi:10.1016/j.tree.2004.12.004. PMID:16701357.

Matsumoto, W.M., and Kazama, T.K. 1974. Occurrence of youngbillfishes in the central Pacific Ocean. In Proceedings of theInternational Billfish Symposium Kailua-Kona, Hawaii, 9–12 Au-gust 1972. Part 2. Edited by R.S. Shomura and F. Williams.National Marine Fisheries Service, Seattle, Wash. NOAA Tech.Rep. NMFS SSRF-675. pp. 238–247.

McDowell, J.R., and Graves, J.E. 2008. Population structure ofstriped marlin (Kajikia audax) in the Pacific Ocean based onanalysis of microsatellite and mitochondrial DNA. Can. J. Fish.Aquat. Sci. 65(7): 1307–1320. doi:10.1139/F08-054.

Mejuto, J., Garcia-Cortes, B., and De La Serna, J.M. 2002.Preliminary scientific estimations of billfish (Family Istiophoridae)landed by the Spanish surface longline fleet targeting swordfish inthe Atlantic Ocean and Mediterranean Sea: years 1999–2000. Col.Vol. Sci. Pap. ICCAT, 54: 826–833.

Morrow, J.E. 1957. Shore and pelagic fishes from Peru, with newrecords and the description of a new species of Sphoeroides. Bull.Bingham Oceanogr. Collect. 16: 5–55.

Nakamura, I. 1985. FAO species catalogue. Vol. 5. Billfishes of theworld. An annotated and illustrated catalogue of marlins,

sailfishes, spearfishes and swordfishes known to date. FAO Fish.Synop. 125.

Nei, M., and Chesser, R.K. 1983. Estimation of fixation indices andgene diversities. Ann. Hum. Genet. 47(3): 253–259. doi:10.1111/j.1469-1809.1983.tb00993.x. PMID:6614868.

Nishikawa, Y., Kikawa, S., Honma, M., and Ueyanagi, S. 1978.Distribution atlas of larval tunas, billfishes and related species:results of larval surveys by R/V Shunyo Maru, and Shoyo Maru,1956–1975. Far Seas Fisheries Research Laboratory, Shimizu,Japan. pp. 1–100.

O’Reilly, P.T., Canino, M.F., Bailey, K.M., and Bentzen, P. 2004.Inverse relationship between FST and microsatellite polymorph-ism in the marine fish, walleye pollack (Theragra chalco-gramma): implications for resolving weak population structure.Mol. Ecol. 13(7): 1799–1814. doi:10.1111/j.1365-294X.2004.02214.x. PMID:15189204.

Palumbi, S.R., Martin, A.P., Romano, S., McMillan, W.O., Stice, L.,and Grabowski, G. 1991. The simple fool’s guide to PCR. Spec.Publ. Department of Zoology, University of Hawaii, Honolulu,Hawaii.

Pritchard, J.K., Stephens, M., and Donnelly, P. 2000. Inference ofpopulation structure using multilocus genotype data. Genetics,155(2): 945–959. PMID:10835412.

Purcell, C.M. 2009. Genetic analysis of population structure instriped marlin, Tetrapturus audax, in the Pacific Ocean. Ph.D.dissertation, Department of Biological Sciences, University ofSouthern California, Los Angeles, Calif.

Purcell, C.M., Harrison, J.S., and Edmands, S. 2009. Isolation andcharacterization of 10 polymorphic microsatellite markers fromstriped marlin, Tetrapturus audax. Mol. Ecol. Resour. 9(6): 1556–1559. doi:10.1111/j.1755-0998.2009.02709.x. PMID:21564958.

Raymond, M., and Rousset, F. 1995. GENEPOP 3.3: population geneticsoftware for exact test and ecumenism. J. Hered. 86: 248–249.

Rooker, J.R., Alvarado Bremer, J.R., Block, B.A., Dewar, H., DeMetrio, G., Corriero, A., Kraus, R.T., Prince, E.D., Rodriguez-Marin, E., and Secor, D.H. 2007. Life history and stock structureof Atlantic bluefin tuna (Thunnus thynnus). Rev. Fish. Sci. 15(4):265–310. doi:10.1080/10641260701484135.

Rozas, J., Sánchez-Delbarrio, J.C., Messeguer, X., and Rozas, R.2003. DnaSP, DNA polymorphism analyses by the coalescent andother methods. Bioinformatics, 19(18): 2496–2497. doi:10.1093/bioinformatics/btg359. PMID:14668244.

Saenz-Agudelo, P., Jones, G.P., Thorrold, S.R., and Planes, S. 2009.Estimating connectivity in marine populations: an empiricalevaluation of assignment tests and parentage analysis underdifferent gene flow scenarios. Mol. Ecol. 18(8): 1765–1776.doi:10.1111/j.1365-294X.2009.04109.x. PMID:19243510.

Seutin, G., White, B.N., and Boag, P.T. 1991. Preservation of avianblood and tissue samples for DNA analysis. Can. J. Zool. 69(1):82–90. doi:10.1139/z91-013.

Shomura, R.S. 1980. Summary report of the billfish stock assessmentworkshop Pacific resources. US Dept. Comm. NOAA-TM-NMFS-SWFC-5.

Squire, J. 1972. Catch distribution and related sea surface temperaturefor striped marlin (Tetrapturus audax) caught off San Diego,California. Proceedings of the International Billfish Symposium1972. NOAA Tech. Rep. NMFS SSRF-675 pp. 188–193.

Squire, J., and Suzuki, Z. 1990. Migration trends of striped marlin(Tetrapturus audax) in the Pacific Ocean. Planning the future ofbillfish research in the 90s and beyond. Part II. Contributed papers.Edited by R.H. Stroud. National Coalition for Marine Conserva-tion, Savannah, Ga. pp. 67–80.

Tamura, K., Dudley, J., Nei, M., and Kumar, S. 2007. MEGA4:Molecular Evolutionary Genetics Analysis (MEGA) software



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

version 4.0. Mol. Biol. Evol. 24(8): 1596–1599. doi:10.1093/molbev/msm092. PMID:17488738.

Theisen, T.C., Bowen, B.W., Lanier, W., and Baldwin, J.D. 2008.High connectivity on a global scale in the pelagic wahoo,Acanthocybium solandri (tuna family Scombridae). Mol. Ecol.17(19): 4233–4247. doi:10.1111/j.1365-294X.2008.03913.x.PMID:19378403.

Ward, R.D., Woodwark, M., and Skibinski, D.O.F. 1994. Acomparison of genetic diversity levels in marine, freshwater and

anadromous fishes. J. Fish Biol. 44(2): 213–232. doi:10.1111/j.1095-8649.1994.tb01200.x.

Weir, B.S., and Cockerham, C.C. 1984. Estimating F-statistics for theanalysis of population structure. Evolution, 38(6): 1358–1370.doi:10.2307/2408641.

Worm, B., Sandow, M., Oschlies, A., Lotze, H.K., and Myers, R.A.2005. Global patterns of predator diversity in the open oceans.Science, 309(5739): 1365–1369. doi:10.1126/science.1113399.PMID:16051749.



Can

. J. F

ish.

Aqu

at. S

ci. D

ownl

oade

d fr

om w

ww

.nrc

rese

arch

pres

s.co

m b

y U

NIV

OF

SOU

TH

ER

N C

AL

IFO

RN

IA o

n 10

/25/

11Fo

r pe

rson

al u

se o

nly.

Documents

Resolving the genetic structure of striped marlin, Kajikia audax, in … · 2013. 11. 26. · Resolving the genetic structure of striped marlin, Kajikia audax, in the Pacific Ocean